News

AI Assistants for Logistics: Making Institutional Knowledge Operational in Live Environments

19.06.2026

Over the course of many long-running logistics projects, a second layer of the system gradually emerges. It is often critical to daily operations, yet rarely visible at first glance: the accumulated technical knowledge surrounding processes, interfaces, custom logic, and customer-specific enhancements.

What can initially be traced through a handful of core project documents becomes distributed over time across multiple documentation versions, technical specifications, interface descriptions, change requests, and implementation records. While the system continues to evolve, so does the volume of information that must later be brought back into context for informed decision-making.

This challenge typically becomes apparent when new requirements arise or when technical questions emerge regarding functionality that was implemented years earlier. In most cases, the necessary information exists and is often documented in great detail. The real effort lies in reconnecting that knowledge across project phases, document versions, and file types. Before any meaningful evaluation can begin, teams often spend considerable time simply locating and reconstructing the relevant information.

This exact situation recently surfaced in a project for a long-standing retail customer of EPG. While commissioning a new warehouse facility that included additional system integrations and an expanded control scope, the project team recognized that although the existing documentation landscape was comprehensive, it had become increasingly difficult to navigate for operational questions. Special processes, historical enhancements, and customer-specific developments had been documented over many years, but the information was scattered across numerous sources. Adding to the complexity, much of this knowledge was not organized in a consistent structure and was spread across different document types, versions, and formats.

When Documentation Needs to Be Understood, Not Just Searched

To make this accumulated knowledge available again in a way that directly supports ongoing project work, key project documents were integrated into EVA.

EVA is the AI Assistant of AURA, EPG’s AI-native Supply Chain Execution environment. EVA unlocks documented expertise from multiple sources, connects related information, and makes it usable within the context of specific business and technical questions. The objective is not simply to search documents faster. The real value lies in making technical information available within its operational context.

EVA links process descriptions, technical specifications, interface documentation, and recorded system changes to reconstruct a complete and reliable understanding of the system. Unstructured information from different sources is brought together into a coherent technical context, making it easier to answer concrete project-related questions.

For this project, the integrated knowledge base included:

• Current functional specification documents
• Process documentation for existing warehouse operations
• Technical descriptions of automation integrations
• Interface documentation from multiple project phases
• Additional project documentation
• Historical change requests and documented enhancements

Based on these sources, a tailored solution was created that makes existing project knowledge available for targeted technical analysis and decision-making.

Initial Applications Within the Ongoing Project

The first tests deliberately focused on questions that regularly occur in day-to-day project work. The primary objective was not the development of new functionality, but rather the understanding and assessment of existing system logic. Which parameter settings are currently active? What functional differences exist between standard functionality and customer-specific enhancements? Which historical change explains a process exception that is still visible today?

In mature installations, answering these questions is often time-consuming because the relevant information rarely resides in a single document. Enhancements may have been implemented over many years, documented in different project phases, and described within the context of the specific requirements that existed at the time. Using EVA, these relationships could be established significantly faster. Existing configurations could be explained based on available documentation, technical dependencies became visible, and historical developments could be reconstructed and understood in a structured manner.

Evaluating New Requirements More Effectively

The approach became particularly valuable when a new business requirement had to be assessed during the ongoing project. The focus was the introduction of Multi-Order Picking within existing warehouse processes. The challenge was not simply defining the new functionality. The real complexity lay in understanding how the requirement would interact with an already evolved system landscape. The project team therefore first analyzed the existing LFS structure within the affected process area. From there, they evaluated which requirements were already covered by existing functionality, where process adjustments would be necessary, and where additional discussions or technical clarification would be required.

Within a short period of time, a reliable foundation for further evaluation was established. Particularly valuable was the fact that technical relationships no longer had to be manually reconstructed from multiple sources, as they were already available in a structured format through the existing knowledge base.

Where EVA Supports Project Teams in Practice

The benefits become particularly evident in environments where systems have evolved over many years and where relevant knowledge is documented but not immediately accessible in daily operations.

Why the Greatest Benefits Emerge in Existing Installations

The longer a system has been in operation, the more the nature of project work changes. At a certain point, implementing new requirements is no longer the primary challenge. The greater effort lies in understanding and evaluating the existing environment with sufficient confidence before any changes are made. This becomes particularly apparent when teams need to answer questions such as:

• Which previous project decision influences the current system logic?
• Why was a specific enhancement originally introduced?
• Which dependencies must be considered before making a change?
• What differences exist between documented standard functionality and actual customer-specific processes?

In these situations, project teams benefit significantly when existing knowledge is not merely archived, but becomes immediately accessible within its proper technical context.

Conclusion: Documentation Becomes Active Project Knowledge Again

The experience gained so far demonstrates that EVA’s value grows directly with the quality and depth of the underlying knowledge base. The more comprehensive the available project documentation, technical specifications, and historical change records, the more accurately technical relationships can be reconstructed and understood.

For existing installations in particular, this creates a fundamentally different approach to managing institutional project knowledge. Documentation is no longer limited to serving as a reference archive. Instead, it becomes an active component of technical project work and operational decision-making.

Next Steps: Systematically Unlocking Institutional Knowledge

The next phase focuses on further expanding the existing knowledge base of the assistant. Additional project documentation, supplementary interface specifications, and historical project records will be integrated incrementally. The objective is not simply to collect more data, but to increase technical depth in those areas where project teams most frequently require reliable answers and informed evaluations. The key questions are therefore:

• Which knowledge domains currently generate the highest research effort?
• Which technical relationships should be available more quickly during project execution?
• Where does fragmented documentation create unnecessary coordination overhead?

If you would like to learn how institutional project knowledge can be systematically unlocked within your organization and how EVA can support this process, talk to us. Our experts can help you transform existing documentation into an active and immediately usable source of project